Statalist


[Date Prev][Date Next][Thread Prev][Thread Next][Date index][Thread index]

Re: st: -nlogit- question


From   rgates@stata.com (Richard Gates)
To   statalist@hsphsun2.harvard.edu
Subject   Re: st: -nlogit- question
Date   Fri, 29 Feb 2008 09:59:26 -0600

Jared Gross is inquiring about the use of -nlogit-: 
 
> I'm trying to use the nlogit command in Stata 10 to estimate a
> degenerate model of mergers and acquisitions.  Unfortunately, 
> I keep getting the error:
> "no cases remain after removing invalid observations."
 
> My data looks like the following:
> 
> Ticker Year Covariates GoesPrivate PublicTakeover Event ContinuesIndep
> A	 1999	x		0		0	0	1	
> A	 2000	x		0		1	1	0		
> B	 1998	y		0		0	0	1		
> B	 1999	z		0		0	0	1
> B	 2000	z		1		0	2	0
> 
> 
> And the model I want to estimate is:
> 
> Company either (1) has  ContinuesIndependently == 1 OR (2) is subject
> to Event == 1 | Event == 2
> 
> If Event == 1 | Event == 2, then either (A) GoesPrivate == 1 OR (B)
> PublicTakeover == 1, so that I have a nested model in which the first
> logit determines whether a company continues <in my data> to the next 
> year or exits through "Takeover" (i.e. Event == 1 | Event == 2); and 
> then a second, degenerate stage in which the company, if Takeover == 1, 
> either GoesPrivate == 1 or PublicTakeover == 1.
> 
> How do I estimate this?  I tried to use the following commands:
> 
> 
> . egen id = group(ticker)
> . nlogitgen takeoverType = Event(Cont: 0, takeover: 1|2)
> . nlogit GoesPrivate covariates || takeoverType: other_covariates,
> +  base(cont) ||, case(id)

The alternatives variable should be the Event variable and it should 
be specified after the last || in the -nlogit- model specification.  I 
believe the hierarchy should look like the following

   takeovertype       Event
   ------------------------------
    takeover |------  2 (private)
             |------  1 (public)

        Cont  ------  0 (degenerate)


For each case coded in the -id- variable one event must occur which is
coded in the dependent variable as an indicator.  The dependent variable
specified in this model is -GoesPrivate- an indicator for Event==2 (I
think) so this does not look valid.

I am thinking the model specification should look something like:

. nlogit choice covariates || takeoverType: other_covariates,
>  base(Cont) || Event:, case(id)

There must be another variable in his data set that indicates which Event
did occur.

I hope this helps.

-Rich
rgates@stata.com
*
*   For searches and help try:
*   http://www.stata.com/support/faqs/res/findit.html
*   http://www.stata.com/support/statalist/faq
*   http://www.ats.ucla.edu/stat/stata/



© Copyright 1996–2014 StataCorp LP   |   Terms of use   |   Privacy   |   Contact us   |   What's new   |   Site index